FLOOD09620 - Laserfiche WebLink

Home Browse Search

I , I t ! f I ~ , I , I JARRETI AND TOMUNSON: REGIONAL INTERDISClPUNARY PALEOF1.00D METHOD rivers in the study area are relatively armored with cobQ!e and boulders and floodplain sediments typically are fine-grained but stable for long periods (Table 2), paleoflood reconstruc- tions reflect reiatively stable conditions. The HWM-PSI rela- tions developed from recent floods in the western United States [Jarrett ef al., 1996J, which included several documented 1995 floods in northwestern Colorado, help to reduce the un- certainty of paleodischarge estimates. In addition, when using paleoflood techniques to estimate peak discharge of recent large floods where gaged flood data were available to assess the reliabiiity, the paleoflood estimates were within about 10% of large gaged floods and further document the value of the critical-depth method (Tabie 2). Therefore paleoflood esti- mates for this study are believed to have total uncertainties of about 25 to 30%. The greatest sources of uncertainty on flood variability are natural or anthropogenic climate change (variability) effects. Paleoflood estimates incorporate the effects of climatic changes on hydrology during the period of the paleoflood record [Jarrett, 1991J. Certainly, moderate climate changes (or other changes such as wildfire effects on flooding or vegetation changes) have occurred during the Holocene; however, these effects are reflected in the maximum flood preserved at a site. Paleoflood data where the maximum age during which the flood occurred is at ieast 5000 years are denoted with large, soiid triangles, and small, solid triangles denote a maximum age of less than 5000 years (Figure 9). The envelope curve of maximum flooding incorporating the paieoflood data (Figure 9) is about 20 to 25% larger than contemporary maximum flooding in about the past 100 years since streamflow monitor- ing began (Figure 9). This modest increase likeiy is due to the large spatial extent of the database and relatively iow- magnitude flooding in northwestern Colorado. Variability in climate and basin conditions during the Holocene does not appear to have had a large impact on flood magnitude, and the assumption of stationarity may be valid for the upper end of the flood-frequency CUIVes in the study area. Thus the enve- lope curve (Figure 9) probably reflects an upper bound of flooding during the Holocene in northwestern Colorado. More quantification (e.g., using one-dimensional or two- dimensional hydraulic modeling to calculate paleoflood dis- charges, using absolute-age dating of flood deposits, more ro- bust flood-frequency parameter estimation procedures, regional flood-frequency analysis with paleoflood data, etc.) would improve the accuracy of individual paleoflood estimates and better quantification of uncertainties. However, the inter- pretation that no substantial flooding has occurred during the Holocene in northwestern Colorado, including Elkhead Creek, would not differ. While use of complex procedures might pro- vide a more precise quantitative description of the data, dis- charge and frequency estimates of extreme floods in a basin may be readily estimated by the paleoflood techniques de- scribed above that provide a cost-effective approach. A critical assumption for calculation of PMP estimates is geographic transposition of storm events from geographically and climatologically similar locations to watershed of interest. However, the NRC [1994] cautions that storm transposition and moisture ma'cimization need to be for a slightly different location in the same climatic region. Regional analyses of rain- fa]], streamflow, and paleoflood data in the present study pro- vide information to evaluate the assumptions about large rain- storms in northwestern Colorado. The assumption that large rainstorms or rain on snow pro- 2979 duce iarge floods in the Rocky Mountains [FEMA, 1976; Han- sen ef aI., 1977, 1988] has implications for dam safety and floodplain management. Although a number of streamflow- gaging stations in the Yampa River basin had over 75 years of record, but no large rainfall floods, these iong-term gaged data were assumed not to be representative of extreme flood po- . tential from rainfall by FEMA (1976J. Thus the flood hydrology for some studies was based on transposing distant, large rain- storms from Arizona, New Mexico, and southwestern Colo- rado into northwestern Colorado and using rainfall-runoff modeling to adjust the upper end of the gaged flood-frequency relation [FEMA, 1976J. The flood.frequency relation for Elk- head Creek at Elkhead Reservoir developed by Ayres Associ- ates, Inc. (written communication, 1996) (Figure 11 and Tabie 3) essentia]]y is the same as the flood-frequency relations from this study up to about the 20-year flood. The Ayres relation sharpiy increases above the 20-year flood, falls outside the confidence iimits of the regional flood-frequency reiations above the 50-year flood, exceeds the maximum paleoflood for the basin at a recurrence interval of about 150 years, and exceeds the envelope curve value of 250 m3 s-\ which is not reasonabie hydrologically. Similar to Eikhead Creek, the FEMA and gaged flood- frequency relations for the Yampa River at Steamboat Springs (Table 3 and Figure 12), where data collection began in 1904, have good agreement to about the 50-year flood. For larger recurrence intervals the FEMA reiation increases sharply and does not fall within the 95% confidence limits for the flood- frequency relation based on streamflow data. In addition, the FEMA 500-year flood is almost double the maximum paleo- flood estimate of 311 m' S-I. The PMF for Stagecoach Res- ervoir iocated on the Yampa River upstream from Steamboat Springs (Figure 12) aiso far exceeds a 10,OOO-year recurrence interval. Similar results for Walton Creek near Steamboat Springs are iisted in Table 3. The difference for larger recurrence intervals primarily re- sults from transposition of distant rainstorms over basins in northwestern Colorado and then using rainfall-runoff model- ing to estimate the upper end of flood-frequency relation as well as the PMF. The gage and paieoflood data provide infor- mation that can be used to refine assumptions used to estimate extreme flooding using storm transposition and rainfall-runoff modeling to at ieast a recurrence interval of 5000 years. The paleoflood data provide no support for sharp upward slope increase of the frequency curve. To help place the flood and paleoflood data in a regional probabilistic context, the EMA relations (with averagel regional skew) for the eight stations (Tabie 3) were plotted versus drainage area (Figure 13). Although flooding results from several factors (basin siope, precipitation indices, vege- tation, etc.) other than drainage area, there is a fairiy good relation between gaged sites. In addition, the envelope curve defined by the paleoflood data also can be placed in a proba- bility context. The site-specific PMP study conducted for the Elkhead Creek drainage basin west of the Continental Divide in north- western Colorado revisited various issues related to the PMP under the explicit conditions which exist at Elkhead Reservoir and other reservoirs in northern Colorado (Tomlinson and Solak, 1997J. These issues included a physical accounting of the effect of topography on storm transpositioning, downslope wind flows under PMP storm conditions, and high-altitude moisture depletion. The combined results of the hydrologic il '1 ,: I , , i )1' ~i ,t' ~.. ~ 'I ,. " 'Z ~j ~:: " ,. .:~ 'il